C-Reactive Protein Concentrations Are Very High and ... - CiteSeerX

Papers in Press. Published December 12, 2008 as doi:10.1373/clinchem.2008.115360 The latest version is at http://www.clinchem.org/cgi/doi/10.1373/clinchem.2008.115360 Clinical Chemistry 55:2 000 – 000 (2009)

Lipids, Lipoproteins, and Cardiovascular Risk Factors

C-Reactive Protein Concentrations Are Very High and More Stable Over Time Than the Traditional Vascular Risk Factors Total Cholesterol and Systolic Blood Pressure in an Australian Aboriginal Cohort Tomer Shemesh,1,2,3 Kevin G. Rowley,3,4* Alicia J. Jenkins,3 James D. Best,3 and Kerin O’Dea3,5

BACKGROUND: Stability of circulating high-sensitivity C-reactive protein (hsCRP) concentrations has implications for its utility in assessing cardiovascular disease (CVD) risk. We sought to determine hsCRP reproducibility in an indigenous Australian cohort with a view to use hsCRP as a marker of future CVD in communitybased risk-factor screenings. METHODS: Seventy people living in a community on the northern coast of Australia participated in 2 risk-factor screenings over a median (interquartile range) follow-up time of 829 (814 –1001) days. hsCRP was measured by high-sensitivity nephelometry. RESULTS:

Geometric mean hsCRP concentrations at baseline and follow-up were 4.5 and 5.1 mg/L, respectively (P ⫽ 0.220), and Pearson product-moment correlation was 0.775. The proportion of people at high CVD risk (hsCRP ⬎3.0 mg/L) at baseline was 67.1% and remained consistently high (68.6%) at follow-up. Linear regression analysis for follow-up hsCRP as a function of baseline hsCRP, sex, and differences in total and regional body fatness showed that baseline hsCRP was the single predictor in the model, accounting for 63.9% of the total variance in follow-up hsCRP (Pmodel ⬍ 0.001). Prevalence agreement (95% CI) between baseline and follow-up for the hsCRP ⬎3.0 mg/L category was 84% (73%–92%) (PMcNemar ⫽ not significant), and ␬ coefficient was fair (0.64, compared with 0.31 for systolic blood pressure ⱖ140 mmHg and 0.43 for total cholesterol ⱖ5.5 mmol/L).

1

Menzies School of Health Research, John Mathews Building, Royal Darwin Hospital, Darwin, NT, Australia; 2 Institute of Advanced Studies, Charles Darwin University, Darwin, NT, Australia; 3 The University of Melbourne, Department of Medicine, St Vincent’s Hospital, Fitzroy, VIC, Australia; 4 Onemda VicHealth Koori Health Unit, Centre for Health and Society, School of Population Health, The University of Melbourne, Parkville, VIC, Australia; 5 The University of Melbourne, Department of Medicine, St Vincent’s Hospital, Fitzroy, VIC, Australia.

CONCLUSIONS:

hsCRP concentrations remained consistently reproducible over time across a wide concentration range in an Aboriginal cohort. Correlations between concentrations over time were better than for other traditional CVD risk factors. hsCRP concentration has potential as a marker of future CVD risk.

© 2008 American Association for Clinical Chemistry

The pentraxin C-reactive protein (CRP)6 is an acutephase reactant that is mainly produced in the liver in response to infection or injury. Recently, low concentrations of C-reactive proteins (measured using highly sensitive assays and thus abbreviated to hsCRP) have emerged as a potentially useful marker of cardiovascular disease (CVD) risk (1 ). The American Heart Association (AHA) has advocated for its use as a predictor of CVD in clinical practice (2 ). High premature CVD mortality is an important contributor to the 20-year shorter median life expectancy of indigenous Australians relative to nonindigenous Australians (3 ). We have previously reported high hsCRP concentrations in Australian Aboriginal populations (4 ), and we hypothesized that chronic inflammation may be an important contributor to the excess CVD risk (5 ). The aim of this study was to determine the degree of reproducibility of hsCRP concentrations over time in an Aboriginal cohort, to determine if subclinical elevation of hsCRP is a chronic condition in this population, with a view to using hsCRP as a marker of CVD risk in community-based risk-factor screenings.

* Address correspondence to this author at: Onemda VicHealth Koori Health Unit, Centre for Health and Society, School of Population Health, The University of Melbourne, Parkville, VIC 3010, Australia. Received July 28, 2008; accepted November 18, 2008. Previously published online at DOI: 10.1373/clinchem.2008.115360 6 Nonstandard abbreviations: CRP, C-reactive protein; hsCRP, high-sensitivity CRP; CVD, cardiovascular disease; AHA, American Heart Association.

1

Copyright (C) 2008 by The American Association for Clinical Chemistry

Research Design and Methods The research design and methods are reported in detail elsewhere (4 ). Briefly, we screened 70 participants living in an isolated coastal Aboriginal community in the northeast of Arnhem Land in 2 risk-factor screening programs between 2001 and 2004. The median time (interquartile range) between the first and second assessment of hsCRP concentrations in this cohort was 829 (814 –1001) days. Venous blood samples were collected and kept cold until centrifuged, and serum was stored under liquid nitrogen until transferred to storage at ⫺80 °C. We measured serum hsCRP using a commercial, high-sensitivity nephelometric assay (BN-II Nephelometer; Dade Behring Diagnostics), with a lower limit of detection of 0.175 mg/L. Inhouse intra- and interassay CVs for hsCRP at 2.4 mg/L were 1.9% and 4.3%, respectively. We measured cholesterol in lithium heparin plasma, glucose in fluoridated plasma, and hemoglobin A1c in EDTA-treated whole blood. Lipid and glucose analyses were performed by routine methods in a diagnostic laboratory accredited by the National Association of Testing Authorities (SouthPath Flinders Medical Centre, SA Australia). We have not reported estimates of LDL cholesterol concentration because the Friedewald equation used to calculate LDL cholesterol has not been validated for use in Aboriginal populations, which are characterized by very low HDL cholesterol concentrations, increased triglycerides, and small, dense LDL particles (6 ). The Friedewald equation is based on assumptions about the distribution of triglycerides across lipoprotein fractions that may not hold in this context (7 ). Statistical analyses were performed using SPSS (ver. 15.0). We used paired-test to compare baseline and follow-up measurements and a multivariate linear regression model to determine follow-up hsCRP as a function of baseline hsCRP, sex, and change in total and regional adiposity. We used Pearson correlation and Bland–Altman (the relationship of the difference between the 2 measurements to the mean of the 2 measurements (8 )) scatterplots to visually illustrate the relationship between follow-up and baseline hsCRP measurements. For categorical analyses of hsCRP and traditional CVD risk factors, we assessed agreement between paired results using the ␬ statistic. ␬ Coefficient ⬍0.40 was considered poor agreement, 0.40 –0.75 fair agreement, and ⬎0.75 excellent agreement (9 ). We used the McNemar test to assess the level of agreement in prevalence of categories at baseline and follow-up screenings. Statistical significance was accepted at P ⬍ 0.05. 2

Clinical Chemistry 55:2 (2009)

Fig. 1. hsCRP distribution at baseline and follow-up screenings.

Results Two discrepant values (⬎3 SDs of the mean difference between baseline and follow-up hsCRP) were excluded from analyses of hsCRP as a continuous variable but not from categorical analyses. Fig. 1 shows the distribution of hsCRP in the population at baseline and follow-up screenings: no bimodality was detected for hsCRP distribution in the population. Table 1 shows clinical characteristics of the participants: there were trends toward increases in all biochemical and anthropometric measurements, but only the increases in hemoglobin A1c and hsCRP were statistically significant. Importantly, hsCRP concentrations were very high in the cohort in both baseline and follow-up screenings (the median hsCRP concentration for adults in the Busselton Health Survey, which is representative of the general Australian population, was about 1.4 mg/L; Hung et al. (10 )). The correlation between baseline and follow-up hsCRP concentrations was greater than those for glucose, blood pressure, and cholesterol (Table 1), but not greater than for HDL cholesterol, hemoglobin A1c, or body mass index. Fig. 2 shows the relationship between baseline and follow-up hsCRP measurements using Pearson correlation and Bland–Altman scatterplots: Pearson product-moment correlation for baseline and follow-up screenings for (log-transformed) hsCRP was 0.775 (Fig. 2A). Bland– Altman analysis showed a modest increase in hsCRP concentrations over time in the population (Fig. 2B). Difference between baseline and follow-up concentrations was not associated with mean CRP (regression coefficient ⫽ 0.188, P ⫽ 0.876). Linear regression analysis for follow-up hsCRP as a function of baseline hsCRP, sex, and differences in total and regional body fatness showed that baseline hsCRP was the single significant predictor in the model [P ⬍ 0.001, accounting

hsCRP Reproducibility in an Aboriginal Cohort

Table 1. Characteristics of participants in baseline and follow-up screenings (n ⴝ 70). Screeninga Rb

Baseline

Follow-up

Geometric mean hsCRP, mg/L (95% CI)

0.775

4.6 (3.5–5.8)

5.1 (3.9–6.7)

0.220

Total cholesterol, mmol/L

0.731

4.7 (1.0)

4.8 (0.9)

0.148

Triglycerides, mmol/L

0.539

2.08 (1.32)

2.30 (1.34)

0.176

HDL cholesterol, mmol/L

0.825

0.95 (0.27)

0.96 (0.25)

0.521

Hemoglobin A1c, %

0.815

6.1 (1.6)

6.5 (1.9)

0.003

Fasting plasma glucose, mmol/L

0.541

6.5 (3.3)

6.8 (4.0)

0.443

2

Body mass index, kg/m

0.964

26.1 (5.5)

26.1 (5.4)

0.942

Waist circumference, cm

0.722

96 (13)

96 (18)

0.998

Weight, kg

0.969

74 (16)

75 (16)

0.201

Systolic blood pressure, mmHg

0.582

119 (20)

119 (22)

0.902

Variable

P valuec

Diastolic blood pressure, mmHg

0.610

71 (10)

71 (12)

0.987

Pulse pressure, mmHg

0.556

48 (14)

49 (14)

0.846

a

Data are mean (SD) unless noted otherwise. Pearson product-moment correlation, baseline vs follow-up. c Paired t-test. b

for 63.9% (Pmodel ⬍ 0.001) of the total variance in hsCRP at follow-up]. Finally, we tested the prevalence agreement between baseline and follow-up hsCRP measurements across AHA hsCRP risk categories for CVD and hsCRP ⬎10.0 mg/L, shown in Table 2: 67% and 69% of the cohort were at high risk (hsCRP ⬎3.0 mg/L) of CVD at baseline and follow-up, respectively. Additionally, there were more cases with hsCRP ⬎10.0 mg/L during follow-up (nbaseline ⫽ 17 vs nfollow-up ⫽ 21), but this increase was not statistically significant (PMcNemar ⫽ 0.424). The proportions of all hsCRP-risk categories did not change significantly over time (all PMcNemar ⫽ not significant), and agreement for placing individuals within categories was fair at hsCRP ⬎10.0 mg/L (␬ ⫽ 0.50, P ⬍ 0.001) and higher for hsCRP ⬎3.0 mg/L (␬ ⫽ 0.64, P ⬍ 0.001). For comparison, agreement (␬ statistics) for the conventional CVD risk factors total cholesterol (ⱖ5.5 mmol/L) and systolic blood pressure (ⱖ140 mmHg) in this population were lower than for hsCRP ⬎3.0 mg/L (Table 2). ␬ Statistics for low HDL cholesterol, overweight, and hyperglycemia were greater than for hsCRP ⬎3.0 mg/L. Discussion Circulating hsCRP concentrations were highly reproducible in this cohort at the population level, even after a median interval of ⬎2 years between the 2 measurements. Reproducibility of hsCRP concentrations in this population indicates that the first measurement of

hsCRP reflects a chronic inflammatory condition. The proportion of the cohort in different hsCRP-risk categories did not change significantly over the median period of 829 days, indicating that the CVD risk profile remained consistent at the population level. As expected, within-person hsCRP concentrations varied somewhat more across all risk categories as indicated by ␬ statistics. However, there was good agreement with respect to individuals having hsCRP ⬎3.0 mg/L, that is, falling within the AHA high-risk category on both occasions. The observed within-person variability was actually lower when compared with that of the conventional traditional CVD risk factors total cholesterol and blood pressure, whether considered as continuous (Table 1) or binary (Table 2) variables. In support, Ockene et al. (11 ) reported that ⬎60% of hsCRP concentrations in apparently healthy adults in the US (mean hsCRP 2.8 mg/L) remained in the same quartile in a repeat measurement after 12 months. Furthermore, Ridker et al. (12 ) reported strong correlation (r ⫽ 0.60, P ⬍ 0.001) of circulating hsCRP concentrations after 60 months of follow-up in post–myocardial infarction patients. In the British Regional Heart Study, underestimation of CVD risk due to time-related variability in risk factors was similar for CRP and systolic blood pressure (but slightly lower than for total cholesterol) [Emberson]. Thus hsCRP measurements at the population and individual levels in this Aboriginal population are stable and comparable with other traditional risk factors. Clinical Chemistry 55:2 (2009)

3

Fig. 2. (A) Scatterplot for the relationship between baseline and follow-up hsCRP. (B) Bland–Altman plot of difference in hsCRP reading (follow-up minus baseline) against mean of hsCRP readings. Solid line represents mean difference, dotted lines represent limits of agreement (2 SDs of mean difference).

The AHA has suggested repeating measures of hsCRP ⬎10.0 mg/L within 2 weeks to exclude an acute inflammatory condition at the time of testing (2 ). The exclusion of samples with hsCRP ⬎10.0 mg/L has been widely accepted in cross-sectional studies, including studies on indigenous populations (13, 14 ). In the present study, no bimodality for hsCRP distribution was evident in either baseline or follow-up. These findings are in agreement with other large-scale studies, such as the Women’s Health Study (15 ), where the risk of the first CVD event increased linearly across increasing deciles of hsCRP, even after stratifying by the 4


Framingham risk score and controlling for diabetes, with no evidence of any threshold effect. Moreover, the authors also proposed that people with hsCRP ⬎10.0 mg/L should be regarded at very high risk of CVD, recommending another hsCRP-risk category for CVD. Taken together, there was no firm evidence for a threshold effect at hsCRP 10.0 mg/L, prompting a revision of its use in clinical settings. Other groups with increased circulating hsCRP in association with chronic inflammatory conditions (rheumatoid arthritis and systemic lupus erythematosus) also experience greatly increased CVD risk and shorter life expectancy (16, 17 ). That hsCRP concentrations were highly reproducible over time is consistent with other published data on inflammation, diet, infections, and abnormal metabolism. In the Reykjavik prospective study, the correlation of hsCRP concentrations at a 12-year interval was 0.59, similar to blood pressure and total cholesterol in that study (18 ). Rowley et al. (5 ) reported very high hsCRP concentrations in 171 participants in a chronic disease study in an Aboriginal community in Western Australia, in association with high levels of vascular cell adhesion molecules (a measure of vascular inflammation) and an inverse correlation with plasma carotenoids. McDonald et al. (19 ) reported high hsCRP concentrations in association with high infectious load in another Australian Aboriginal community. Additionally, results from the Strong Heart Study of CVD in 3277 participants from American Indian communities (13 ) showed a median (interquartile range) hsCRP of 3.2 (1.8 –5.1) mg/L with a relatively high proportion of people with hsCRP ⬎10.0 mg/L (16% vs 24% at baseline in this cohort), consistent with abnormally high hsCRP concentrations in economically disadvantaged populations. The stability of hsCRP after multiple freeze-thaw cycles and a lengthy period in storage has been reported (20, 21 ) and was also reproduced in our laboratory (data not shown). Technically, a number of assays can be used to determine hsCRP concentrations in blood. High levels of agreement among commercially available hsCRP measurement assays in categorizing patients into CVD-risk categories has been demonstrated (18, 22 ), indicating that such assays are comparable for clinical purposes. Physiologically, hsCRP is unaffected by diurnal variation and has a relatively long half-life (23 ). An additional advantage is that, like total and HDL cholesterol, hsCRP can be measured in nonfasting blood samples. In conclusion, hsCRP appears to be useful for monitoring CVD risk, and possibly evaluating clinical and public health interventions in indigenous communities with high levels of cardiovascular disease risk.

hsCRP Reproducibility in an Aboriginal Cohort

Table 2. Prevalence and agreement between baseline and follow-up CRP-risk categories and CVD risk factors. Baseline, % (n)

Variable

Follow-up, % (n)

Agreement, % (95% CI)a

␬ (P)

McNemar P

hsCRP, mg/L ⬍1.0

a

7.1 (5)

8.6 (6)

87 (76–94)

0.11 (0.343)

1.000

1.0–3.0

25.7 (18)

22.9 (16)

93 (83–97)

0.38 (0.001)

0.804

3.0–10.0

42.9 (30)

38.6 (27)

67 (55–78)

0.32 (0.007)

0.678

⬎10.0

24.3 (17)

30.0 (21)

80 (68–88)

0.50 (⬍0.001)

0.424

⬎3.0

67.1 (47)

68.6 (48)

84 (73–92)

0.64 (⬍0.001)

1.000

Systolic blood pressure ⱖ140 mmHg

18.6 (13)

10.1 (7)

83 (71–90)

0.31 (0.006)

0.146

Total cholesterol ⱖ5.5 mmol/L

17.1 (12)

16.2 (11)

84 (72–91)

0.43 (⬍0.001)

1.000

HDL cholesterol ⬍1.03 (male)/⬍1.29 (female) mmol/L

81.4 (57)

76.5 (52)

93 (83–97)

0.78 (⬍0.001)

0.375

Body mass index ⱖ25.0 kg/m2

55.7 (39)

55.7 (39)

94 (89–99)

0.88 (⬍0.001)

1.000

Fasting plasma glucose ⱖ6.1 mmol/L

22.9 (16)

34.3 (24)

86 (75–93)

0.66 (⬍0.001)

0.021

Proportion of the total sample falling within the same category on both occasions.

Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article. Authors’ Disclosures of Potential Conflicts of Interest: Upon manuscript submission, all authors completed the Disclosures of Potential Conflict of Interest form. Potential conflicts of interest: Employment or Leadership: None declared. Consultant or Advisory Role: None declared. Stock Ownership: None declared. Honoraria: None declared.

Research Funding: This work was funded by grants from the National Health and Medical Research Council of Australia (#124319) and the Novo Nordisk Regional Diabetes Support Scheme. K.G. Rowley was supported by a VicHealth Public Health Research Fellowship and an NHMRC Senior Research Fellowship. Expert Testimony: None declared. Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation or approval of manuscript. Acknowledgments: The authors thank all participants and the community involved in the screening and the Community Council for their support of the work, Steve Farish for his statistical advice, and Connie Karschimkus for her technical support.

References 1. Verma S, Szmitko P, Ridker P. C-reactive protein coming of age. Nature (Lond) Clin Pract Cardiovasc Med 2005;2:29 –36. 2. Pearson TA, Mensah GA, Alexander RW, Anderson JL, Cannon RO I, Criqui M, et al. Markers of inflammation and cardiovascular disease: application to clinical and public health practice: a statement for healthcare professionals from the Centers for Disease Control and Prevention and the American Heart Association. Circulation 2003;107:499 –511. 3. Trewin D, Madden R. The health and welfare of Australia’s Aboriginal & Torres Strait Islander peoples. Canberra: Australian Institute of Health and Welfare; 2005. 4. Shemesh T, Rowley KG, Jenkins A, Brimblecombe J, Best JD, O’Dea K. Differential association of C-reactive protein with adiposity in men and women in an Aboriginal community in northeast Arnhem Land of Australia. Int J Obes 2007;31: 103– 8. 5. Rowley K, Walker KZ, Cohen J, Jenkins AJ, O’Neal D, Su Q, et al. Inflammation and vascular en-

6.

7.

8.

9. 10.

dothelial activation in an Aboriginal population: relationships to coronary disease risk factors and nutritional markers. Med J Aust 2003;178:495–500. O’Neal DN, Piers LS, Iser DM, Rowley KG, Jenkins AJ, Best JD, et al. Australian Aboriginal people and Torres Strait Islanders have an atherogenic lipid profile that is characterised by low HDLcholesterol level and small LDL particles. Atherosclerosis 2008 Apr 6 [Epub ahead of print]. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 1972;18: 499 –502. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307–10. Rosner B. Fundamentals of biostatistics. Boston: PWS-Kent Publishing, 1990. Hung J, Knuiman MW, Divitini ML, Davis T, Beilby JP. Prevalence and risk factor correlates of elevated C-reactive protein in an adult Australian population. Am J Cardiol 2008;101:193– 8.

11. Ockene IS, Matthews CE, Rifai N, Ridker PM, Reed G, Stanek E. Variability and classification accuracy of serial high-sensitivity C-reactive protein measurements in healthy adults. Clin Chem 2001;47:444 –50. 12. Ridker PM, Rifai N, Pfeffer MA, Sacks F, Braunwald E. Long-term effects of Pravastatin on plasma concentration of C-reactive protein. Circulation 1999;100:230 –5. 13. Best LG, Zhang Y, Lee ET, Yeh J-L, Cowan L, Palmieri V, et al. C-reactive protein as a predictor of cardiovascular risk in a population with a high prevalence of diabetes: the Strong Heart Study. Circulation 2005;112:1289 –95. 14. Anand SS, Razak F, Yi Q, Davis B, Jacobs R, Vuksan V, et al. C-reactive protein as a screening test for cardiovascular risk in a multiethnic population. Arterioscler Thromb Vasc Biol 2004;24: 1509 –15. 15. Ridker PM, Cook N. Clinical usefulness of very high and very low levels of C-reactive protein across the full range of Framingham risk scores. Circulation 2004;109:1955–9.


5

16. Wong M, Toh L, Wilson A, Rowley K, Karschimkus C, Prior D, et al. Reduced arterial elasticity in rheumatoid arthritis and the relationship to vascular disease risk factors and inflammation. Arthritis Rheum 2003;48:81–9. 17. Lee AB, Godfrey T, Rowley KG, Karschimkus CS, Dragicevic G, Romas E, et al. Traditional risk factor assessment does not capture the extent of cardiovascular risk in SLE. Int Med J 2006;36:237– 43. 18. Rifai N, Tracy RP, Ridker PM. Clinical efficacy of an automated high-sensitivity C-reactive protein assay. Clin Chem 1999;45:2136 – 41.

6


19. McDonald S, Maguire G, Duarte N, Wang XL, Hoy W. C-reactive protein, cardiovascular risk, and renal disease in a remote Australian Aboriginal community. Clin Sci (Lond) 2004;106: 121– 8. 20. Aziz N, Fahey JL, Detels R, Butch AW. Analytical performance of a highly sensitive C-reactive protein-based immunoassay and the effects of laboratory variables on levels of protein in blood. Clin Diag Lab Immunol 2003;10:652–7. 21. Ledue TB, Rifai N. Preanalytic and analytic sources of variations in C-reactive protein measurement:

implications for cardiovascular disease risk assessment. Clin Chem 2003;49:1258 –71. 22. Roberts WL, Moulton L, Law TC, Farrow G, Cooper-Anderson M, Savory J, et al. Evaluation of nine automated high-sensitivity C-reactive protein methods: implications for clinical and epidemiological applications. Part 2. Clin Chem 2001; 47:418 –25. 23. Vingushin DM, Pepys MB, Hawkins PN. Metabolic and scintigraphic studies of radioiodinated human C-reactive protein in health and disease. J Clin Invest 1993;91:1351–7.